another case of USB HDD bridge failure

[goontron]

This one is a 3tb Toshiba dwc130 Canvio brought to me because Windows Vista would not "see" it.
It seemed to have issues spinning up and loading the heads, so i checked the power brick which checked fine, next i pulled the drive and swapped it into one of the hotswap bays of my system and it worked fine.
The data was corrupted beyond all recognition so i went on to the drive bridge where i found two dry 220uF 16V CrapXon caps (pix before cap test).
Going into the S.M.A.R.T data i found the drive was ran over temp (63c) for 720 hours, all others are fine with an exception of the number of interface CRC errors.

[kc8adu]

Re: another case of USB HDD bridge failure

some of those usb-sata boards have encryption.
could explain the corruption.
crapxons strike again!
thats a buck to make +5.
noise on +5 can corrupt data too.
not the first toshiba external hdd i have seen die this way.shame on you toshiba!

[goontron]

Re: another case of USB HDD bridge failure

*UPDATE*

the drive is absolutely fine. passed an extended self test and managed to survive a SECURE_ERASE command. it is now my backup drive in my hot swap bay.

[goontron]

Re: another case of USB HDD bridge failure

*UPDATE a year later*
Drive is still going strong, and another SATA to USB bridge failed... This one from a laptop drive enclosure. It has ESATA that is pass-though of the chip, So no pics.

[fzabkar]

Re: another case of USB HDD bridge failure

In addition to the possibility of encryption, some enclosures (eg WD My Book 3TB+ and Seagate Goflex 3TB+) are configured with 4KB sector sizes rather than 512 bytes. When you remove such a drive from its enclosure, you expose its native 512e sectoring and render the file system inaccessible. Essentially the OS will see a 4Kn file system on a 512e physical drive.

[goontron]

Re: another case of USB HDD bridge failure

In addition to the possibility of encryption, some enclosures (eg WD My Book 3TB+ and Seagate Goflex 3TB+) are configured with 4KB sector sizes rather than 512 bytes. When you remove such a drive from its enclosure, you expose its native 512e sectoring and render the file system inaccessible. Essentially the OS will see a 4Kn file system on a 512e physical drive.

Linux wouldn't care.. It does fine with both, and being an EFI system...

[fzabkar]

Re: another case of USB HDD bridge failure

Linux wouldn't care.. It does fine with both, and being an EFI system...

Linux would have exactly the same problem. It has nothing to do with EFI.

[goontron]

Re: another case of USB HDD bridge failure

Linux would have exactly the same problem. It has nothing to do with EFI.

Explain how? After all, every Linux powered NAS i see uses that format...

[fzabkar]

Re: another case of USB HDD bridge failure

A NAS is a network drive whereas a USB mass storage device or internal HDD is a "local" drive. The OS can access a local drive at sector level, but a network drive can only be accessed at file level. The OS does not see the sector size or even the file system in a NAS. That's why Windows can access the files on a NAS with an embedded Linux ext3 file system.

The problem I am describing is one where the USB-SATA bridge firmware presents a USB mass storage device to the host as a 4Kn device but deals with the physical HDD behind the bridge as a regular 512e drive. The bridge does the 4Kn-to-512e translation internally, and this is transparent to the host. As far as the host is concerned, the drive is a 4Kn device and the file system is formatted with a 4KB sector size. This sector size is recorded in the boot sectors of FAT and NTFS volumes.

Here is a recent thread where a user of a 4TB My Book got into trouble when he discarded his bridge board:

http://forum.hddguru.com/viewtopic.php?t=32420

He did eventually recover his data using TestDisk (a data recovery tool with Linux roots), but I suspect that Linux would not have been able to mount the drive. In any case TestDisk did complain of a sector size mismatch. Neither EaseUS or Windows were able to see the data. The fundamental problem is that the partition table reports the wrong starting sector and wrong number of sectors for each partition.

For example, a 4TB drive with a sector size of 4KB would have approximately 1 billion sectors. Therefore the partition table might define an NTFS volume starting at sector 256 and ending at sector 976,562,500. When this drive is installed in its 4Kn enclosure, the OS will find the boot sector at byte offset 1,048,576 (= 256 sectors x 4096 bytes per sector). However, if the drive is removed from its enclosure, then the OS will look for the boot sector at offset 131,072 (= 256 sectors x 512 bytes per sector). In fact the boot sector will now be located at LBA 2048.

[goontron]

Re: another case of USB HDD bridge failure

Another update: Got a 4TB WD My Book in for case swap (Linux host, EXT3) The system had no issues with the 4kn vs 512e problem... Also got a 500gb freeagent in for data recovery (user supplied a new 3tb hgst external) broken power and USB jack... nothing to look at. Same thing with NTFS... No issues accessing data.

[fzabkar]

Re: another case of USB HDD bridge failure

A 500GB FreeAgent is configured with 512-byte sectors. Seagate only configures their bridge firmware with 4096-byte sectors when the capacity exceeds 2TiB, so no problems are to be expected in this particular case.

A 4TB My Book would normally be configured with 4Kn sectoring. However, WD does have a "quick formatter" tool (a misleading misnomer) which allows the user to reconfigure the bridge firmware for a 512-byte sector size. This change is written to the firmware on the bridge PCB and persists after a power cycle. Of course the drive needs to be repartitioned in GPT mode afterwards.

The term "My Book" might also refer to My Book Live which is a linux based NAS. If this is what you have, then that would explain the standard 512e sectoring. (A NAS is a network drive, not a local drive. Therefore its sector size would be transparent to the host.)

[japlytic]

Re: another case of USB HDD bridge failure

I see a large inductor with an 8-pin SOP IC (most likely to be a DC-DC converter) nearby.
So therefore, these electrolytic capacitors (CapXon) were not up to the task of filtering the output of the onboard DC-DC converter.
For the given capacitance and size, I would replace with solid polymer units.

[goontron]

Re: another case of USB HDD bridge failure

A 500GB FreeAgent is configured with 512-byte sectors. Seagate only configures their bridge firmware with 4096-byte sectors when the capacity exceeds 2TiB, so no problems are to be expected in this particular case.

A 4TB My Book would normally be configured with 4Kn sectoring. However, WD does have a "quick formatter" tool (a misleading misnomer) which allows the user to reconfigure the bridge firmware for a 512-byte sector size. This change is written to the firmware on the bridge PCB and persists after a power cycle. Of course the drive needs to be repartitioned in GPT mode afterwards.

The term "My Book" might also refer to My Book Live which is a linux based NAS. If this is what you have, then that would explain the standard 512e sectoring. (A NAS is a network drive, not a local drive. Therefore its sector size would be transparent to the host.)

Yeah, the FreeAgent was nothing...

Here's the thing with the My Book though (not live, i just checked). It has never seen a Windows machine to do a quick format/use the quick formatter tool.